tjh.dev
1{ lib
2, stdenv
3, buildPythonPackage
4, fetchPypi
5, python
6, pythonOlder
7, cython
8, numpy
9, python-dateutil
10, pytz
11, scipy
12, sqlalchemy
13, tables
14, xlrd
15, xlwt
16# Test inputs
17, glibcLocales
18, hypothesis
19, jinja2
20, pytestCheckHook
21, pytest-xdist
22, pytest-asyncio
23, xlsxwriter
24# Darwin inputs
25, runtimeShell
26, libcxx
27}:
28
29buildPythonPackage rec {
30 pname = "pandas";
31 version = "1.5.3";
32 format = "setuptools";
33 disabled = pythonOlder "3.8";
34
35 src = fetchPypi {
36 inherit pname version;
37 hash = "sha256-dKP9flp+wFLxgyc9x7Cs06hj7fdSD106F2XAT/2zsLE=";
38 };
39
40 nativeBuildInputs = [ cython ];
41
42 buildInputs = lib.optional stdenv.isDarwin libcxx;
43
44 propagatedBuildInputs = [
45 numpy
46 python-dateutil
47 pytz
48 ];
49
50 nativeCheckInputs = [
51 glibcLocales
52 # hypothesis indirectly depends on pandas to build its documentation
53 (hypothesis.override { enableDocumentation = false; })
54 jinja2
55 pytest-asyncio
56 pytest-xdist
57 pytestCheckHook
58 xlsxwriter
59 ];
60
61 # Doesn't work with -Werror,-Wunused-command-line-argument
62 # https://github.com/NixOS/nixpkgs/issues/39687
63 hardeningDisable = lib.optional stdenv.cc.isClang "strictoverflow";
64
65 doCheck = !stdenv.isAarch32 && !stdenv.isAarch64; # upstream doesn't test this architecture
66
67 # don't max out build cores, it breaks tests
68 dontUsePytestXdist = true;
69
70 pytestFlagsArray = [
71 # https://github.com/pandas-dev/pandas/blob/main/test_fast.sh
72 "--skip-db"
73 "--skip-slow"
74 "--skip-network"
75 "-m" "'not single_cpu'"
76 "--numprocesses" "4"
77 ];
78
79 disabledTests = [
80 # Locale-related
81 "test_names"
82 "test_dt_accessor_datetime_name_accessors"
83 "test_datetime_name_accessors"
84 # Disable IO related tests because IO data is no longer distributed
85 "io"
86 # Tries to import from pandas.tests post install
87 "util_in_top_level"
88 # Tries to import compiled C extension locally
89 "test_missing_required_dependency"
90 # AssertionError with 1.2.3
91 "test_from_coo"
92 # AssertionError: No common DType exists for the given inputs
93 "test_comparison_invalid"
94 # AssertionError: Regex pattern '"quotechar" must be string, not int'
95 "python-kwargs2"
96 # Tests for rounding errors and fails if we have better precision
97 # than expected, e.g. on amd64 with FMA or on arm64
98 # https://github.com/pandas-dev/pandas/issues/38921
99 "test_rolling_var_numerical_issues"
100 # Requires mathplotlib
101 "test_subset_for_boolean_cols"
102 # DeprecationWarning from numpy
103 "test_sort_values_sparse_no_warning"
104 ] ++ lib.optionals stdenv.isDarwin [
105 "test_locale"
106 "test_clipboard"
107 # ValueError: cannot reindex on an axis with duplicate labels
108 #
109 # Attempts to reproduce this problem outside of Hydra failed.
110 "test_reindex_timestamp_with_fold"
111 ];
112
113 # Tests have relative paths, and need to reference compiled C extensions
114 # so change directory where `import .test` is able to be resolved
115 preCheck = ''
116 cd $out/${python.sitePackages}/pandas
117 export LC_ALL="en_US.UTF-8"
118 PYTHONPATH=$out/${python.sitePackages}:$PYTHONPATH
119 ''
120 # TODO: Get locale and clipboard support working on darwin.
121 # Until then we disable the tests.
122 + lib.optionalString stdenv.isDarwin ''
123 # Fake the impure dependencies pbpaste and pbcopy
124 echo "#!${runtimeShell}" > pbcopy
125 echo "#!${runtimeShell}" > pbpaste
126 chmod a+x pbcopy pbpaste
127 export PATH=$(pwd):$PATH
128 '';
129
130 enableParallelBuilding = true;
131
132 pythonImportsCheck = [ "pandas" ];
133
134 meta = with lib; {
135 # https://github.com/pandas-dev/pandas/issues/14866
136 # pandas devs are no longer testing i686 so safer to assume it's broken
137 broken = stdenv.isi686;
138 homepage = "https://pandas.pydata.org/";
139 changelog = "https://pandas.pydata.org/docs/whatsnew/index.html";
140 description = "Python Data Analysis Library";
141 license = licenses.bsd3;
142 maintainers = with maintainers; [ raskin fridh knedlsepp ];
143 platforms = platforms.unix;
144 };
145}